Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle 10g RAC performance

Re: Oracle 10g RAC performance

From: joel garry <>
Date: 30 Apr 2007 13:21:38 -0700
Message-ID: <>

On Apr 28, 5:42 pm, DA Morgan <> wrote:
> Mladen Gogala wrote:
> > On Sun, 29 Apr 2007 01:10:30 +0200, The Boss wrote:
> >> There will be some companies moving from Oracle to PostgreSQL, others
> >> will start with PostgreSQL and afterwards maybe will come to the
> >> conclusion they need Oracle (or DB2 or whatever). In both cases they
> >> might have a need for someone with skills in both databases.
> > Basically, what I am complaining about is the tendency to push RAC to
> > everybody. What people don't understand is that adding machines also
> > adds a layer of complexity.
> I really don't think our fellow DBAs are that dense. You don't give
> enough credit where credit is due. I think everyone understands that
> there is an added layer, clusterware, of technology. And possibly two
> if you throw in ASM.

I don't think they are that dense either, but from what I've seen they often
have the idea that since Oracle is pushing RAC to everybody, it must be
a good thing. And I do get the feeling people underestimate the additional
risk more layers of complexity adds. People seem to think each layer's
problems are additive or even independent, when they are actually multiplicative.

> But the decisions are not being made by DBAs they are being made by
> CTOs. And the fault, if I can find any, is that those CTOs are willing
> to invest in the RAC licenses but not in training their team so that
> they will be successful.

I'm sure Oracle would be quite willing to certify their team as professionals :-O

Regardless, Mladen is right: RAC is pushed to everybody.

> > I have a lot of experience with Oracle and
> > I needed a lot of help from Gopal when I did my first Oracle 10G/ASM
> > installation. I am very grateful for his help and his book has been of
> > immense help, but RAC is a very complex product. Essentially, RAC
> > makes a good DBA a must.
> You will get no argument from me on that. Nor would I argue that
> training, or a mentor, are essential for a successful design and
> deployment. But that has been true of every major technology advance
> I have witnessed in the last 30 years.

The real question is, how many of those advances are pushed to those who really don't need them? (And of course, there is a corollary question of those who stall at particular technologies, only to need to
catch up later, we certainly see that a lot in this ng).

> Of course there are self-taught DBAs that are very good. But lets be
> honest and acknowledge the majority couldn't describe how an Oracle
> transaction works if given a snap test.

Not to mention if they can describe how a snap works if given a transaction
test :-)

> > Very few small companies are willing to hire
> > a full time DBA just to manage a database.
> Our here in Seattle that is not true but it may be the case where you
> are located. Here if they are willing to step up to the plate for
> Oracle ... they are willing to step up to the plate for a DBA or 500.

This kind of caught my attention: my observation (Southern Californai)
jibes with Mladen's. I couldn't help but wonder if the Seattle area is
different because any smaller shop would likely be MS...

> That said there is still a problem with their willingness to invest in
> that person's training though some companies out here are extremely good
> at taking care of their people.
> We have companies in Seattle and Portland that, literally, sent their
> entire DBA team to take Jonathan Lewis' classes when he has been here.
> is extremely supportive of their employees and has gone so
> far as to contract the University of Washington to teach classes for
> them.
> > In addition to that, I once worked for a company that went from Oracle to
> > Postgres, just because they felt that Oracle is too big, too complex and
> > too expensive for them.
> No doubt. I know companies that have gone bankrupt making bad decisions.
> One or ten or fifty examples does not make a trend.
> > Wang Trading was a small hedge fund in Norwalk,
> > CT that got rid of both Oracle and me. That sort of things tends to catch
> > my attention. Now, they're out of business, I have no idea why.
> I could hazard a guess ... and I will ... bad decision making. A better
> management team likely would have kept both Oracle and you.

Another Wang once was a big computer company. Must be a jinx there. I guess the moral must be if you have a wang, keep it private.

K Gopalakrishnan:



-- is bogus.
And the award for "the largest human alteration of the Earth's
surface." goes to...
Received on Mon Apr 30 2007 - 15:21:38 CDT

Original text of this message